The present invention relates to an audio signal processing method and apparatus which carry out characteristic correction in case of reproducing an audio signal from a speaker device and more particularly relates to a technology preferably to be applied in case of using a speaker device for HIFI reproduction by which reproduction of a high-quality sound is possible.
In the past, various kinds of constitutions were in practical use as a speaker device for HIFI reproduction by which reproduction of a high-quality sound is possible. For example, there is known a speaker device having a three-way constitution in which reproduction bands of the audio signal are divided into three bands of a low band, a middle band and a high band and individual speaker units are provided for respective bands thereof. Reproduction faithful to an input audio signal from a low band to a high band becomes possible in the speaker device having a three-way constitution by using units as speaker units for respective bands in which reproduction characteristics thereof are favorable in respective bands and generally, the reproduction characteristic thereof becomes favorable as compared with a so-called full-range type speaker unit which outputs audio of all bands by a single speaker unit.
Also, other than the constitution in which the reproduced sound of the speaker device is made to be a high-quality sound by adopting such a three-way constitution or a two-way constitution, there has been adopted a constitution in which the characteristic of the audio signal itself supplied to the speaker device is corrected on the side of an amplifier device which is an audio signal processing apparatus such that the audio characteristic outputted from the speaker device is to be improve accordingly. For example, there is a case in which a correction referred to as a loudness control is carried out by an audio amplifier device which performs a processing of amplification of an audio signal driving the speaker device or the like. This loudness control is a control for carrying out a correction process which strengthens a bass portion and a treble portion in the output level thereof as compared with a midrange portion such that a phenomenon that the bass and treble portions sound insufficiently mainly on an occasion of a small volume is to be corrected.
In Jap. Laid-Open Publication No. 2002-171589, there is a description with respect to one example of a reproducing constitution in case of carrying out a loudness correction.
However, the loudness controlled reproduced sound simply strengthens a signal in a specific frequency band approximately uniformly regardless of its level, so that it cannot be said in the strict sense of the word that a faithful reproduction with respect to the input audio signal is achieved and a development of a speaker device capable of achieving a more faithful reproduction with respect to the input audio signal has been desired. More specifically, since the reproduced sound which was loudness-controlled according to a conventional way strengthens a sound which cannot be caught easily on an occasion of a small volume to be reproduced, the bass portion and the treble portion become audible easily as compared with a reproduced sound which is not loudness-controlled and there is an effect of improving the sound quality to a certain degree, but the signal in a specific frequency band is to be strengthened uniformly regardless of a small level of a large level, so that it might happen that it may strengthen also with respect to a signal component which is unnecessary to be strengthened and as a result there is a case in which an unnatural reproduced sound is obtained.
Here, it will be explained with respect to a problem of the reproduced sound in a speaker device of related art, wherein there is a problem of a signal having a small amplitude as an example in a case when the reproduced sound does not reproduce the input audio signal faithfully. More specifically, for example, as shown in FIG. 1A, a case is assumed in which an input audio signal S1 of a continuous waveform having a waveform of a relatively large amplitude and a waveform of a relatively small amplitude. At that time, as a waveform of an audio signal S2 outputted from the speaker, it becomes approximately comparable with the input signal S1 with respect to a waveform of a relatively large amplitude and with respect to a waveform of a relatively small amplitude, there is a trend such that the amplitude thereof becomes smaller than that of the input signal S1. This is because the reproducing characteristic of a signal having small amplitude of a small volume is bad in a speaker unit having a shape provided with a general diaphragm capable of outputting in a relatively large sound and linearity of the input-output characteristic of a small volume signal cannot be assured.
Similarly as shown, for example, in FIG. 1B, when an input audio signal S3 having a waveform of a relatively large amplitude and an input audio signal S4 having a waveform of a relatively small amplitude overlap in time, an audio signal S5 composed by both the signals S3 and S4 is expected to be outputted primarily, but it becomes a state in which an output audio signal S6 having a waveform which is lowered in level as compared with the waveform of that composed signal S5 is to be outputted from the speaker. In a case, for example, when sounds of various musical instruments are to be reproduced concurrently such as a symphony as an audio to be reproduced from the speaker, such an output state may happen.
Further, as shown, for example, in FIG. 1C, in a case when there is an impulse signal as an input audio signal S7 in which a signal amplitude of a specific single frequency lowers gradually, it is true with respect to the waveform of an output audio signal S8 from the speaker that a following characteristic becomes deteriorated more as the level thereof becomes lower.
In any one of the examples of FIGS. 1A to 1c, the output level the signal having amplitude of a small volume becomes smaller than the input signal level with respect to the output from the speaker and it becomes a state in which linearity of a small signal cannot be maintained. When frequency-analyzing the state shown in FIGS. 1A to 1c, it becomes a state shown, for example in FIG. 2. The example of FIG. 2 is an example in which sensitivity is analyzed with respect to a fundamental wave f1 and its harmonics f 2 and f 3 which are higher harmonic waves of the fundamental wave. With respect to the fundamental wave f1 having a high level, it is outputted by a level as it was, but with respect to the harmonics f 2 and f 3 having smaller levels than the fundamental wave, the output sensitivities thereof become as shown by solid lines which are lowered than the primarily expected levels shown by dotted lines.
FIGS. 3A and 3B are drawings showing output characteristics from a low band to a high band in signal levels of a plurality of steps, wherein FIG. 3A shows an ideal characteristic and FIG. 3B is a drawing showing an output characteristic of an actual speaker. As shown in FIG. 3A, it is assumed in an ideal state such that four levels L1, L2, L3 and L4 were spaced approximately equally and it was a flat characteristic from a low band to a high band. At that time, with respect to the levels L1, L2 and L3 having high output levels for the output characteristic of an actual speaker shown in FIG. 3B, output characteristics approximately comparable with the ideal characteristic can be assured, but with respect to the characteristic of the lowest level L4, the levels are to be lowered from the primarily necessary levels by sensitivity a in any frequency bands.
The input-output characteristic diagram of FIG. 4 is a drawing when such sensitivity lowering is seen as a specific frequency characteristic. As shown in FIG. 4, while it is necessary primarily that the output level increased linearly with respect to the increase of the input signal level to the speaker so as to obtain a characteristic x of a dotted line, actually, the level changes approximately linearly in a level of a certain degree or more, but motion of the diaphragm with respect to the input is bad in a specific level or less such that a curved characteristic y is obtained in which the output sensitivity with respect to the input is very bad.
Specifically, in case of, for example, assuming that the maximum level for listening by a general speaker is to be 70 to 100 dBspl (sound pressure level), it can be said that a signal which is lowered from the maximum level by −30 dB to −60 dB does not output a sound volume which is correctly lowered by −30 dB to −60 dB with respect to the maximum level (is not proportional). Tentatively, when assuming a reproduction by a sound volume in which the output of the amplifier device is lowered from 100 dBspl by an amount of 50 dBspl, a sound volume before and after 50 dBspl should be obtained under an ordinary circumstance, but it happens actually, for example, that only an output of 40 dBspl which is lower than that by 10 dBspl can be obtained. In other words, it was recognized by an analysis of the present inventor that linearity cannot be fulfilled precisely and it becomes one of big causes for a phenomenon that a satisfied sound quality cannot be obtained.
It should be noted that the characteristic explained so far is the characteristic in case of a speaker unit having a relatively large diaphragm capable of outputting a relatively large sound volume, but conversely, in case of a speaker unit having a small sized and lightweight diaphragm which is produced for a small volume output such as, for example, a speaker unit for a headphone, there also exists a constitution in which linearity of the input-output is maintained relatively favorably on an occasion of a small volume and linearity of the input-output characteristic cannot be maintained on an occasion of a large sound volume.